models intro

What role do models play within agents?

They are the reasoning engine of agents: they drive the agent's decision-making process; determining which tools to call; how to interpret results; and when to provide a final answer.

What two ways can models be used in LangChain?

1) With agents — dynamically specified when creating an agent. 2) Standalone — called directly outside the agent loop for tasks like text generation; classification; or extraction.

What is the easiest way to initialize a standalone model in LangChain?

Use `init_chat_model` to initialize a model from a chat model provider of your choice.

Why do new model names work immediately without a LangChain update?

Because provider packages pass model names directly to the provider's API.

What format lets you specify both a model and its provider in a single argument?

The `'provider:model'` format; e.g. `'openai:o1'`.

What does the `temperature` parameter control on a chat model?

It controls the randomness of the model's output — higher values make responses more creative; lower values make them more deterministic.

What does the `max_tokens` parameter control?

It limits the total number of tokens in the response; effectively controlling how long the output can be.

What does the `timeout` parameter control?

The maximum time (in seconds) to wait for a response from the model before canceling the request.

What is the default value of `max_retries`; and what does it represent?

Default is 6 — the maximum number of attempts to resend a failed request (network timeouts; rate limits).

What backoff strategy do retries use?

Exponential backoff with jitter.

Which types of errors are retried automatically?

Network errors; rate limits (429); and server errors (5xx).

Which types of errors are NOT retried automatically?

Client errors such as 401 (unauthorized) or 404.

For long-running agent tasks on unreliable networks; what value is suggested for `max_retries`?

Increasing it to 10–15.

What can you do with a chat model instance after configuring `max_retries` and `timeout`?

Pass that instance to `create_agent`; `create_deep_agent`; or call it standalone.

python import os from langchain.chat_models import ________ os.________['OPENAI_API_KEY'] = 'sk-.....' model = init_chat_model('gpt-4o-mini') response = model.________('Why do parrots talk')

```python import os from langchain.chat_models import init_chat_model os.environ['OPENAI_API_KEY'] = 'sk-.....' model = init_chat_model('gpt-4o-mini') response = model.invoke('Why do parrots talk') ```

python model = init_chat_model( 'claude-sonnet-4-6'; # Kwargs passed to the model: ________=0.7; timeout=30; ________=1000; max_retries=6; # Default; increase for unreliable networks )

```python model = init_chat_model( 'claude-sonnet-4-6'; # Kwargs passed to the model: temperature=0.7; timeout=30; max_tokens=1000; max_retries=6; # Default; increase for unreliable networks ) ```

python from langchain.chat_models import init_chat_model model = ________( 'google_genai:gemini-3.6-flash'; ________=10; # Increase for unreliable networks (default: 6) timeout=120; # Seconds; increase for slow connections )

```python from langchain.chat_models import init_chat_model model = init_chat_model( 'google_genai:gemini-3.6-flash'; max_retries=10; # Increase for unreliable networks (default: 6) timeout=120; # Seconds; increase for slow connections ) ```

python os.environ['OPENAI_API_KEY'] = 'sk-.....' ________ = init_chat_model('gpt-4o-mini') response = model.invoke('Why do parrots talk')

```python os.environ['OPENAI_API_KEY'] = 'sk-.....' model = init_chat_model('gpt-4o-mini') response = model.invoke('Why do parrots talk') ```

python model = ________( 'claude-sonnet-4-6'; # Kwargs passed to the model: temperature=0.7; timeout=30; max_tokens=1000; max_retries=6; # Default; increase for unreliable networks )

```python model = init_chat_model( 'claude-sonnet-4-6'; # Kwargs passed to the model: temperature=0.7; timeout=30; max_tokens=1000; max_retries=6; # Default; increase for unreliable networks ) ```